Table of Contents
I have 5 SaaS ideas I want to build right now
Every few days I come up with an idea and add it to this list I have in Apple Notes titled Upcoming builds.
Right now the list looks like this.
![notion image](https://cdn.feather.blog?src=https%3A%2F%2Fusenotioncms.com%2Fproxy-prod%2Fblock%2Fdc1253ba-d690-4cb5-9536-80551ee4ab4f%2F27a79174-129c-4f22-85b8-c69dbd38aa46%2F8198cf18-9aba-4431-b375-3baf7ab45b76%2FScreenshot_2024-06-14_at_12.21.43.png&optimizer=image&quality=80&width=280)
I know, I know. Five upcoming builds? Am I crazy?
Quite possibly.
But my logic is this:
I’m going to use my Bullet Launch boilerplate template for all of them.
So with each build I will be testing the template and finding bugs and improvements to make, which will benefit everyone who has bought the template (or will buy it in future).
Each build should be relatively quick thanks to the template.
And each new launch will help to generate some buzz around it.
So, in theory, it’s all part of a big marketing flywheel.
But a little voice is telling me I’ve invented this whole flywheel just so I can build more silly products nobody will use. 😄
I guess we’ll find out!
My first build: A.I. and voice, an exploding trend
The AI voice calling space is currently blowing up because the technology is just getting to a point where it’s usable.
The idea that soon we’ll be spending time on the phone to A.I. bots instead of humans doesn’t exactly fill me with joy.
But there are plenty of situations where it could improve accessibility of services or actually create value for both companies and their service users.
So I decided to build an AI voice app. Hence “Robot Voice” on my list above (this is a placeholder name, don’t worry!).
What does Robot Voice do?
So far I’ve built in the ability to:
- create an AI agent
- assign it a “mission”
- choose a list of target clients
- schedule calls
The agent will then make outbound phone calls to all the clients and attempt to complete its mission.
For example, it might be calling around suppliers to get a quote on various items.
It then returns a transcript of the call which can be read by AI and summarised (I haven’t built this part yet).
![The “Create a job” workflow in my Robot Voice MVP](https://cdn.feather.blog?src=https%3A%2F%2Fusenotioncms.com%2Fproxy-prod%2Fblock%2F969c7481-d567-485e-b2ae-7552e9d9457c%2F27a79174-129c-4f22-85b8-c69dbd38aa46%2Fd875c8ce-3687-43d2-91e0-98d85c65e1bb%2FXnapper-2024-06-14-12.34.42.png&optimizer=image&quality=80&width=280)
To build it I’ve used Bullet Launch combined with Retell.ai, an AI voice API provider.
So far the build has been fairly smooth although the Retell API is a bit strange in the way it’s structured, and this part has taken me the most time.
The main downside is the cost of the calls.
When you add up all the elements it can be $0.15 a minute which doesn’t sound much but I’m used to testing and tweaking things a million times and that adds up quite fast!
It’s been a very interesting learning experience anyway, and each new API I work with builds my knowledge and confidence.